Skip to main content

The Archive Team QAUDIO Rescue (WARC Collection)

First you need to understand that blind computer users don't use computers in the way normal sighted people do. We use special software called a screen reader which tells us what's on the screen and provides easier access to programs and websites. On Windows, the most popular screen readers are Jaws for Windows, Window-Eyes, System Access, and Non-visual Desktop Access. Because we don't look at the screen, we also use the keyboard exclusively. Screen readers and other programs often make use of unusual keyboard commands and modifiers in order to keep from interfering with whatever program the user might be trying to access.

One of these specialized programs designed for the blind was a Twitter client written by Christopher Toth (Twitter username @Mongoose_Q, but lots of us just call him Q.) This client was called Qwitter, and it ran on Windows. It interface d directly with popular screen readers. What made it so easy to use and popular is that many of its functions could be performed without switching out of whatever application you were currently in. Did you just get a new tweet? Assuming you were in your home buffer, Ctrl+Windows+Up arrow and that tweet was read to you. Want to reply? Ctrl+Windows+r, type in your message, hit Enter, and you could go back to whatever you were doing.

In October of 2010, the then current version of Qwitter had built-in integration with for playback of sound clips. It didn't have any way to upload to, though. To do that, you had to go to, sign in, and upload your file. To fix this, Q decided to create his own audio service which he called Q-audio. Since he was the author of Qwitter, he integrated full support of upload and playback to Q-audio right into the client. He also added one other important feature. Right from within any dialog where you were composing a new tweet, mention, or direct message, there was an "attach audio" button. Within that dialog was a record button. Now for the first time, anyone could record audio right from within the client, play it back to see if it sounded the way they wanted, and if so, upload it to Q-audio. Q-audio would then return the URL to the info page about that file. In those days, the URL took the form "" where XXXXX was the index of that particular file in the Q-audio datastore. Of course you could also go to and upload files from the site, though you couldn't record a brand new file. You could also upload an existing file from within the Qwitter client.

On approx. October 10, 2010, Qwitter 4.5 was released. This version's major new feature was Q-audio integration. The first days of posts show how people who maybe weren't technically-minded enough to run a sound recorder, record a file, save it, then upload it to an audio service could record their voice, their introductions, thoughts, etc. into this new service with very little effort.

A few days later, someone created an Internet radio station called Twadio. This station played files from Q-audio. Realizing the operator was obviously mass-downloading, I asked how he did it. He sent me the scraping script he was using. I immediately began scraping Q-audio on an irregular schedule.

As time went on, people continued to use Q-audio to send messages to each other in their own voice. Some people, of course, decided to use the service to share commercial music with their friends (a blatant copyright infringement.) Sometime in May or June of 2011, to shorten the URLs Q-audio produced, Q decided to convert the service to begin using shorter IDs in the form of a number in base 36. The older IDs were still valid for the older files, so as not to break all existing Q-audio links. At this point my scraping came to an end, since the script I was using couldn't handle the new URL format.

When Qwitter 5.0 was put into public beta, it had a warning about sending Q-audio posts to friends in direct messages, indicating that although the link was only sent to the person you intended, the actual file could be downloaded by anybody who figured out the file's ID or scraped the service.

In 2012 Dreamhost's problems started to become a real pain. Apparently the mere mention of the possible end of Q-audio sometime down the road caused many people to start scraping the service. Q started blocking the IP addresses of scrapers, especially those who used poorly-written scripts. The following tweets illustrate:

Christopher Toth posted Seriously You should improve your script to not continue to download when it keeps getting 404's. #fail. Christopher Toth posted @Mongoose_Q And 403's, now. #CutOff Christopher Toth posted I love how I explicitly ask people not to scrape q-audio. And someone starts... Yep. #winning.

Late on the night of Saturday, April 14, 2012, the beginning of the end came. The following tweets illustrate Q being fed up with Dreamhost:

Christopher Toth posted So the VPS is down. That's fine. But hahahahahahaha BTW the control panel is down, too! #dreamhost #fail #GetMeOutOfHere! Christopher Toth posted Okay. I'm going to be moving off DreamHost. Christopher Toth posted There's nowhere comparably cheap, so this probably means the end of q-audio.

He made it clear that people were not to scrape. Word got around on Twitter that he didn't want scrapers, but if you had any favorite Q-audio files, now would be a good time to download them. He never gave a definite date for the cutoff. I alerted Jason Scott of the pending shutdown, and we talked some on IRC. At this point I started uploading my private archive from my scraping days to an Archive Team server. That upload was completed early Saturday morning, April 21.

On Friday, April 20, 2012, the end of Q-audio came. There was no fanfare, it just suddenly went down, returning 404 errors to all queries. Through the effort of an Archive Team member, we were able to save approximately 98% of the files stored there before it was shut down.

Archive Team